Inventory and monitoring of wine microbial consortia

Appl Microbiol Biotechnol (2007) 75:149–164 DOI 10.1007/s00253-006-0798-3 APPLIED MICROBIAL AND CELL PHYSIOLOGY Inventory and monitoring of wine microbial consortia Vincent Renouf & Olivier Claisse & Aline Lonvaud-Funel Received: 8 September 2006 / Revised: 5 December 2006 / Accepted: 6 December 2006 / Published online: 19 January 2007 # Springer-Verlag 2007 Abstract The evolution of the wine microbial ecosystem is generally restricted to Saccharomyces cerevisiae and Oenococcus oeni, which are the two main agents in the transformation of grape must into wine by acting during alcoholic and malolactic fermentation, respectively. But others species like the yeast Brettanomyces bruxellensis and certain ropy strains of Pediococcus parvulus can spoil the wine. The aim of this study was to address the composition of the system more precisely, identifying other components. The advantages of the polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE) approach to wine microbial ecology studies are illustrated by bacteria and yeast species identification and their monitoring at each stage of wine production. After direct DNA extraction, PCR-DGGE was used to make the most exhaustive possible inventory of bacteria and yeast species found in a wine environment. Phylogenetic neighbor-joining trees were built to illustrate microbial diversity. PCR-DGGE was also combined with population enumeration in selective media to monitor microbial changes at all stages of production. Moreover, enrichment media helped to detect the appearance of spoilage species. The genetic diversity of the wine microbial community and its dynamics during winemaking were also described. Most importantly, our study provides a better understanding of the complexity and diversity of the wine microbial consortium at all stages of the winemaking process: on grape berries, in must during fermentation, and in wine during aging. On grapes, 52 V. Renouf (*) : O. Claisse : A. Lonvaud-Funel Faculté d’œnologie, UMR 1219 Université Bordeaux 2, INRA, ISVV, 351 Cours de la Libération, 33405 Talence, France e-mail: v-renouf@enitab.fr different yeast species and 40 bacteria could be identified. The diversity was dramatically reduced during winemaking then during aging. Yeast and lactic acid bacteria were also isolated from very old vintages. B. bruxellensis and O. oeni were the most frequent. Introduction Winemaking is a complex microbial process in which yeast and bacteria play key roles. After crushing, yeasts, mainly Saccharomyces cerevisiae, consume sugars to produce ethanol during alcoholic fermentation. Subsequently, in some highly acidic red wines, as well as in white wines with aging potential, lactic acid bacteria (LAB), mainly Oenococcus oeni, convert malic acid into lactic acid during malolactic fermentation. Both yeast and bacteria produce aromas responsible for sensorial wine properties. When both types of fermentations are completed, microbial populations must be reduced because post-fermentation microbial metabolisms are prejudicial to the wine’s organoleptic qualities. This is particularly true for volatile phenol synthesis by the yeast Brettanomyces bruxellensis, which confers off-odors to wine (Chatonnet et al. 1992), as well as causing exopolysaccharide (Walling et al. 2001), biogenic amine (Coton et al. 1998), and ethyl carbamate production (Uthurry et al. 2005) by some LAB strains. All of these microbial species are naturally present on grape skins (Renouf et al. 2005a). Winemaking equipment such as tanks and barrels constitute other natural microbial sources. All of that underline how important it is to monitor the microbial presence of grape skins, fermenting must, winery equipment, and wine during aging. Traditional microbiological methods, such as microscopic observation and isolation in selective nutritive media, allow 150 the visualization and enumeration of various microbial populations. These methods have focused on total yeast (TY), non-Saccharomyces yeast (NS), lactic acid bacteria (LAB), and acetic acid bacteria (AAB). More efficient media have been developed to favor the growth of minority species (Renouf and Lonvaud-Funel 2006) because their detection implies the use of a specific enrichment medium. Furthermore, the development of the epifluorescence (Millet and Lonvaud-Funel 2000) method has revealed the existence in wine of viable but noncultivable (VNC) microorganisms with metabolic activities, but are unable to grow in a nutritive medium. Regarding identification, phenotypic tests based on fermentation and assimilation characters, respiratory properties, etc...have enabled the identification of some wine microbial species. With these methods, cultivation is always necessary. However, the real conditions under which most species actually grow in their natural habitat are not always clear. This makes it very difficult to develop universal media for cultivating all species. Therefore, studies relying on culture-dependent tools are likely to lead to biased results based on unrepresentative cultivation conditions. Moreover, these methods are laborious and time-consuming, as the time necessary for colony growth causes an additional delay. The alternative offered by molecular methods is increasingly studied. These methods are independent of gene expression and based solely on similarity and dissimilarity of DNA sequences. In most instances, molecular methods target ribosomal genes (Amann and Ludwig 2000). Polynucleotide probes hybridizing species-specific sequences were developed to identify lactic acid bacteria species in wine (Lonvaud-Funel et al. 1991). However, cross-hybridization was observed between closely related species. Restriction fragment length polymorphism (RFLP) and the specific enzymatic digestion of ribosomal genes have produced specific restriction profiles of wine bacteria (Le Jeune and Lonvaud-Funel 1994) and yeast (Guillamon et al. 1998) species. However, RFLP applications on complex microbial mixtures lead to confusing and noninterpretable electrophoresis profiles because the restriction products of different species may be superimposed. This is also the main drawback of the RAPD (random amplified polymorphic DNA) method, which is based on the use of nonspecific primers and amplification conditions favorable to random amplification of short sequences. These molecular methods overcome the influence of the physiological state, but they remain limited when resolving complex microbial mixtures. The solution resides in the combination of species-specific sequence amplification and their separation according to species–genotype characteristics. Denaturing gradient gel electrophoresis (DGGE) is based on this same principle. The DGGE enables the Appl Microbiol Biotechnol (2007) 75:149–164 separation of polymerase chain reaction (PCR) amplicons of the same size but of different sequences. PCR doublestrand amplicons in the gel are subjected to an increasingly denaturing environment. The migration is stopped when the DNA fragments are complementally denatured. To resolve complex microbial mixtures, PCR-DGGE must target single copies of the species-discriminating gene. Concerning yeast analyses, the regions currently targeted are the D1/D2 domains of the large ribosomal subunit 26S rRNA gene (Kurtzman and Robnett 1997). The 18S rRNA gene has also been tested (Ampe et al. 2001), but bands corresponding to nonmicrobial DNA were observed. With regard to bacteria, the chromosomal regions commonly targeted on the 16S rRNA gene are V1, V3, V6, and V8 (Muyzer et al. 1993). However, in a given species, this gene is present in several copies, and each one can have a different sequence. This is especially prevalent for V1 and V6 (Coenye and Vandamme 2003). Moreover, primers targeting V3, V6, and V8 led to the amplification of yeast (Saccharomyces sp., Candida sp.), mold (Botrytis cinerea, Fusarium laterium; Lopez et al. 2003), and vegetable (Dent et al. 2004) DNA. Investigation of other genes was necessary to overcome the 16S rRNA limitations. Gene used for PCR-DGGE analysis should consist of a mosaic of well-conserved regions that can be used as amplification primers and variables. Among bacterial species, the gene coding for the beta subunit RNA polymerase fits this definition (Renouf et al. 2006a). The primers are chosen by analysis of consensus sequences flanking variable regions. Based on this methodology, Rantsiou et al. (2004) and Renouf et al. (2006b) have, respectively, designed primers targeting rpoB regions for bacteria. These primers amplify sequences of 250 pb. However, Rantsiou et al. (2004) reported that the primer set they used led to co-migration of different species amplicons, especially various Pediococcus species and O. oeni. That is a serious drawback for their use in enology. Last, but not least, undetermined bands are often observed in DGGE gels. They can be isolated from the gels, reamplified using similar conditions and PCR primers, and then sequenced and compared with sequences available in databanks to determine species identity. Therefore, the phylogenetic properties of regions targeted in PCR-DGGE are crucial to bringing undetermined band and determined sequences closer together and to making a more complete species inventory. The objectives of this work were (1) to illustrate the advantages of PCR-DGGE in a systemic approach to wine microbiology, (2) to investigate the diversity of bacteria and yeast species involved in winemaking, (3) to study the dynamics of microbial populations according to different stages of winemaking, and (4) to focus on the detection of spoilage species. Appl Microbiol Biotechnol (2007) 75:149–164 151 Materials and methods Samples Healthy grapes, fermenting musts, and wines were collected from several vineyards in the Bordeaux area: the Libourne region, Pessac-Léognan, and the Medoc. Six different grape varieties were studied: Merlot, Cabernet Sauvignon, Cabernet Franc, and Petit Verdot (red wine varieties) as well as Sémillon and Sauvignon Blanc (white wine varieties) on a total of twenty-four plots located on eight estates. After the harvest, we monitored the microbial species changes throughout the winemaking process for a Sémillon wine and a Merlot wine until bottling. Water used to wash stainless tanks at the end of the fermentation as well as water used to clean barrels during racking after 3 months of aging were also analyzed. Several old and very old vintages of bottled wine, listed in Table 1, were also studied. Each grape, must, wine, and cleaning water sample was analyzed in triplicate. DNA extraction A universal protocol for bacteria and yeast studies and for berry washing solutions, winery equipment cleaning solutions, musts, and wine samples was used. Microbial biomass was collected from 10 ml of fermenting must, 100 ml of wine during aging and in bottle, 100 ml of berry washing solution and winery equipment cleaning solution by centrifugation (30 min, 10,000×g, 4°C), and the pellet was washed in 2 ml of Tris 10 mM (GenApex)–EDTA Table 1 Origins and vintages of the wines analyzed 1 mM (GenApex; TE) buffer. After a second centrifugation (10,000×g for 15 min at 4°C), the supernatant was discarded, and the pellet resuspended in 300 μl of 0.5 mM EDTA pH 8. Three hundred microliters of glass beads were added (∅=0.1 mm), and samples were mixed at maximum speed for 15 min. Then, 500 μl of nuclei lysis (Promega) and 300 μl of protein precipitation solution (Promega) was added and mixed for 30 s. Precipitation of cellular fragments was made on ice for 5 min. After another centrifugation (10,000×g for 5 min at 4°C), the supernatant containing the DNA was transferred to a new microcentrifuge tube. Residual polyphenols were precipitated after addition of 100 μl of a 10% polyvinyl-pyrrolidone (PVP; Sigma-Aldrich) solution and vortexing at high speed for 10 s. For highly tannic wines, this step can be repeated. After centrifugation (10,000×g for 5 min at 4°C), the supernatant was transferred to a 1.5-ml microcentrifuge tube containing 500 μl of isopropanol. The tube was gently mixed by inversion until a visible mass of DNA could be seen and left at −20°C for 3 h. After centrifugation (10,000×g for 20 min at 4°C), 300 μl of a room temperature 70% ethanol solution was added to the pellet before a final centrifugation (10,000×g for 5 min at 4°C). Ethanol was carefully removed and the tube dried. Fifty microliters of PPI (Pour Preparation Injectable) water with 1 μl of RNase (Promega) were used to rehydrate DNA overnight at 4°C. After rehydratation, this DNA was stored at −20°C. This extraction can also be used to process biomasses collected on Petri dishes. The biomass was collected in TE buffer before the first step of centrifugation. DNA concentrations were standardized (50 ng/μl) by measuring optical density at 260 nm with a SmartSpec (+) Bio-Rad spectrophotometer. Wines Appellations Vintages PCR-DGGE conditions I II III IV V VI VII VIII IX X XI XII XIII XIV XV XVI XVII XVIII XIX XX Pessac–Léognan Pauillac Pauillac Pauillac Pauillac Pauillac Pessac–Léognan Pessac–Léognan Margaux Saint-Emilion Margaux Pessac–Léognan Pauillac Pauillac Pauillac Pauillac Saint-Emilion Saint-Emilion Margaux Margaux 1909 1926 1928 1929 1947 1949 1974 1981 1981 1981 1982 1989 1990 1993 1994 1995 1996 1998 1998 2003 PCR-DGGE protocols using NL1/LS2 and rpoB1, rpoB1o/ rpoB2 primers, respectively, described by Cocolin et al. (2000) for yeast and Renouf et al. (2006b) for bacteria, were used with some modifications. The PCR program began with an initial touchdown step in which the annealing temperature was lowered from 59 to 45°C in increments of 1°C every cycle. Furthermore, 20 additional cycles were carried out with an annealing temperature of 45°C. Electrophoresis took place in vertically acrylamide (Promega) gel with chemical denaturing conditions provided by urea (Sigma-Aldrich) and formamide (SigmaAldrich). A solution of 100% chemical denaturant consists of 7-M urea and 40% (v/v) formamide in milliQwater. Ten microliters of PCR amplicons at 50 ng/μl were loaded with high-density marker: a mixture of glycerol (80%) and Tris 10 mM (GenApex)–EDTA 1 mM (GenApex) buffer (20%), and bromophenol blue (J.T. Baker). Based on the Cocolin et al. (2000) protocol for yeast analysis, we used a lower 152 denaturing gradient from 20 to 45% of urea and formamide, instead of 30 to 60%. Based on the observations of Sigler et al. (2004) revealing the importance of the time/voltage ratio of separation sensitivity, we also modified the voltage applied during migration: 200 instead of 120 V, associated with a similar migration time (4 h). These modifications provided a better view of the species and improved separation power in the gel. After migration, gels were stained with SYBR Green I and visualized under UV. Analysis of sequences and species identification When unknown bands appeared in the DGGE gel, small blocks of acrylamide were excised and put in TE buffer overnight at 4°C to allow DNA to diffuse out of the gel. The DNA was then used for re-amplification using primers (without GC clamp) under the same conditions as PCRDGGE. The PCR-amplified products were purified using a Qiaquick® PCR Purification Kit provided by Qiagen and then sequenced. These sequences were compared with those available in the GenBank database and similitude percentages calculated after alignment. We selected the closest referenced sequence for each one, and we built a phylogenetic tree to provide information on the genetic relationship between the sequence analyzed and the referenced species. Phylogenetic and molecular evolutionary analyses were conduced using MEGA version 2.1 (Kumar et al. 2001). Primers were excluded from the analysis. The neighborjoining function (Saitou and Nei 1987) was selected, phylogenetic distance was calculated according to Kimura’s method, and 1,000 repetitions were made for bootstrap (Felsentein 1985). Only significant bootstrap values (above 50%) are indicated on the branches. Appl Microbiol Biotechnol (2007) 75:149–164 was made by cycloheximide addition and the time of incubation. Even some wild Saccharomyces strains may be resistant to 0.1% cycloheximide, they are more sensitive than non-Saccharomyces species, and they are unable to grow in 10 days at 25°C. Concerning non-Saccharomyces, no sensitivity to the cycloheximide was reported among the enology currently identified (Renouf et al. 2005a, 2006c). The cycloheximide concentration chosen was the better compromise between Saccharomyces inhibition and large non-Saccharomyces revelation. Three different bacteria populations (anaerobic Grampositive species, consisting mainly of lactic acid bacteria, anaerobic, and facultative anaerobic Gram-negative species, and aerobic Gram-negative species, mainly composed of AAB) using three different selective nutritive media [GJLAB, GJ-AAB, and ZPP (Coton and Coton 2003)] were also studied. In these media, yeast growth was inhibited by adding pimaricine (Delvocid, DSM Food Specialities). Growth of Gram-positive bacteria was inhibited by the addition of penicillin (Sigma-Aldrich). To obtain solid media, 20 g/l of agar was added to each medium. The enumeration was made on plates with between 30 and 300 colonies. In some cases, the detection of minor species implied the use of a specific enrichment medium, especially in the case of NS and LAB, in looking for such spoilage species as B. bruxellensis and P. parvulus. The composition of the two enrichment media enrichment Brettanomyces bruxellensis (EBB) and enrichment lactic acid bacteria (ELAB) is also listed in Table 2. For analysis of old vintages, we also used epifluorescence as described by Millet and Lonvaud-Funel (2000) to estimate the viability of the residual microorganism. Colony isolation and counts Results For enumeration and isolation purposes, each undiluted sample and dilution series was plated onto different nutritive media selected for specific use and listed in Table 2. The total yeast-yeast peptone glucose (TY-YPG) medium was used for the TY population. Biphenyl (Fluka) and chloramphenicol (Sigma Aldrich), respectively, inhibited mold development and bacterial growth. The addition of 0.1% cycloheximide (Sigma Aldrich) to the NSYPG medium eliminated the Saccharomyces spp. and made it possible to enumerate the NS population (Renouf et al. 2006c). We currently used 0.1% cycloheximide during several investigations, and we had never had a problem with Saccharomyces species resistance. In fact, the incubation lasted 10 days. It corresponded to the delay of inhibition of Saccharomyces species. When incubation time was increased (15, 20 days) some Saccharomyces species could grow. Then, elimination of Saccharomyces species Microbial ecology of grape surfaces Microbial inventories of grape surfaces were made by bacteria and yeast PCR-DGGE analyses during three successive vintages at various estates. In addition to the direct sample analysis, we also incubated the berries for 10 days in EBB and ELAB enrichment media. The number of species increased, including less important species. This is illustrated in Fig. 1, where the detection of O. oeni on grape surfaces was only possible after the grapes were incubated in an ELAB medium. With regard to yeast, 52 different sequences were extracted from DGGE gels obtained during the monitoring of all the plots studied in the eight estates. Some sequences were identical and corresponded to sequences available in the databank; others were compared to the most closely related species (Table 3). This was the same for bacteria and, Targets Yeast Population Total yeast Non-Saccharomyces Name of the medium YPG-TYa YPG-NSa Composition Glucose 20 g/l, yeast extract 10 g/l, bactotryptone 10 g/l, pH=5.0 (orthophosphoric acid) Glucose 20 g/l, yeast extract 10 g/l, bactotryptone 10 g/l, pH=5.0 (orthophosphoric acid) Selective agent(s) Biphenyl: 150 mg/l Biphenyl: 150 mg/l Chloramphenicol: 100 mg/l Incubation conditions 25°C 5 days Aerobic Bacteria Chloramphenicol: 100 mg/l Cycloheximide: 500 mg/l 25°C 10 days Aerobic Anaerobic and facultative anaerobic Gram-positive (mainly composed by lactic acid bacteria) Aerobic Gram-negative (mainly composed by acetic acid bacteria) Anaerobic and facultative anaerobic Gram-negative EBBa GJ-LABa ELABa GJ-AABa ZPPa Yeast extract 1.5 g/l, yeast extract 1.5 g/l, (NH4)2SO4 0.5 g/l, MgSO4 7H20 0.2 g/l, grape juice 200 ml/l, ethanol 40 ml/l, Tween 80 0.5 ml/l, pH=5.0 (orthophosphoric acid) Biphenyl: 200 mg/l Yeast extract, 5 g/l Tween 80 1 ml/l, grape juice 500 ml/l, pH=4.8 (NaOH, 10N) Glucose 10 g/l, peptone 10 g/l, yeast extract 5 g/l, MgSO4 7H2O 200 mg/l, MnSO4 H2O 50 mg/l, grape juice 250 ml/l pH=4.8 (KOH, 10N) Pimaricine: 50 mg/l Yeast extract 5 g/l, Tween 80 1 ml/l, grape juice 500 ml/l, pH=4.8 (NaOH, 10N) Glucose 20 g/l, peptone 5 g/l, yeast extract 3 g/l, malt extract 3 g/l Pimaricine: 100 mg/l Penicillin: 15 mg/l Pimaricine: 100 mg/l Penicillin: 30 mg/l 25°C 5 days Aerobic 25°C 5 days Anaerobic Pimaricine: 100 mg/l Chloramphenicol: 50 mg/l Appl Microbiol Biotechnol (2007) 75:149–164 Table 2 List of the culture media used in this work Cycloheximide: 250 mg/l 25°C 10 days Aerobic 25°C 10 days Anaerobic 25°C 10 days Anaerobic a NS Non-Saccharomyces; EBB Enrichment Brettanomyces bruxellensis; GJ Grape juice; ELAB Enrichment for lactic acid bacteria; LAB Lactic acid bacteria; AAB Acetic acid bacteria; ZPP Zymomonas pimaricine penicillin TY Total yeasts 153 154 Appl Microbiol Biotechnol (2007) 75:149–164 Concerning yeast, Fig. 5 provides a view from the vineyard to bottled Merlot wine. The yeast population increased on berries during ripening. Then, during winemaking, three separate phases could be distinguished. The first phase was AF, when the TY population increased to 108 CFU/ml. S. cerevisiae was the main species. The second was after the first racking, accompanied by the addition of sulfur dioxide, when the yeast population declined. The third and final phase was during barrel aging, when the population once again rose to 103–104 CFU/ml, at which point it stayed stable. The yeast level was contained thanks to the addition of free SO2 (never less than 25 mg/l), repeated racking (every 3 months), and fining. A significant decline of the yeast species diversity was noted from the beginning of fruit set. Then, during aging and up until bottling, there was only one single species, B. bruxellensis, especially for the red grape varieties. Microbial ecology of bottled wine Fig. 1 Example of O. oeni detection from Merlot berries at the berry set after direct incubation in an enrichment medium for lactic acid bacteria (ELAB; II). Direct analysis (I) did not reveal its presence although all the sequences could not be clearly identified, the tree showed great diversity on the grape surface (Fig. 2). The majority of the bacterial groups are present, in particular the proteobacteria, which are not commonly described in enology. Interestingly, the most common enological yeast (S. cerevisiae, B. bruxellensis) and bacteria (O. oeni, P. parvulus, G. oxydans) were detected on grape skins from the first stages of development. Monitoring industrial winemaking Figures 3 and 4 show the combination of the culturedependent method and rpoB PCR-DGGE culture-independent identification for red wine (Merlot) and white wine (Sémillon). The curve of anaerobic Gram-positive population primarily represents the level of lactic acid bacteria. There was greater diversity in white wine fermenting must than in red wine fermenting must in DGGE gels. This diversity remained high in fully fermented white wine, whereas it rapidly decreased in red wine. However in both cases, O. oeni was the LAB species most resistant to AF. In Merlot wine, O. oeni was the only bacteria found during malolactic fermentation, reaching a population of 107 CFU/ml. In Sémillon wine, malolactic fermentation did not take place, and the population never exceeded 104 CFU/ml after alcoholic fermentation, due to the addition of sulfur dioxide. It is important to note a decrease in microbial diversity during the initial stages of winemaking as compared to the species found on grape skins. We analyzed bottled wines many years after they were bottled to estimate the residual microflora. Epifluorescence observations confirmed the enumeration on solid media and PCR-DGGE profiles. For example, in Fig. 6, a B. bruxellensis characteristic cell morphology was revealed by epifluorescence, and a single band could be seen on NL1/LS2 DGGE gel corresponding to B. bruxellensis. This showed the viability and cultivability of the yeast population. Five sequences were obtained for bacteria analysis, and nine different bands were extracted from the DGGE gel and sequenced for yeast analysis. Sequence V was close to basidiomycetous Rhodotorula mucilaginosa, and the eight others corresponded to ascomycetous species. Some sequences were clearly identified: I, III, and IV, respectively, similar to B. bruxellensis, Pichia anomala, and Zygosaccharomyces bailii sequences. The others were closely related to the reference species with differing degrees of similarity. The VII sequence was least similar, but, even so, the rate of similarity was greater than 85%. The accuracy of these comparisons was confirmed by high bootstrap values. O. oeni and P. parvulus species were clearly identified. The frequency of detection for each sequence pattern for all the wine listed in Table 1 is showed in Figs. 7 and 8. For yeast, B. bruxellensis was always detected, and for bacteria, O. oeni was the main species. P. parvulus was detected in half the wines. Microbial ecology of fermentation vat and oak barrel surfaces Figure 9 illustrates the rpoB PCR-DGGE direct analysis of water used for tank cleaning at different stages of winemaking. At the time the wine was run off from vat, O. oeni Appl Microbiol Biotechnol (2007) 75:149–164 155 Table 3 Results of the comparison of the isolated sequences by NL1/LS2 PCR-DGGE analyses of grape berries surface with those present in GenBank from NCBI database Isolate designation Species GenBank accession no. I II III IV V VI VII VIII IX X XI XII XIIII XIV XV XVI XVII XVIII XIX XX XXI XXII XXIII XXIV XXV XXVI XXVII XXVIII XXIX XXX XXXI XXXII XXXIII XXXIV XXXV XXXVI XXXVII XXXVIII XXXIX XXXX XXXXI XXXXII XXXXIII XXXXIV XXXXV XXXXVI XXXXVII XXXXVIII IL L LI LII Rhodotorula glutinis Rhodotorula glutinis Rhodotorula glutinis Rhodosporidium krachilovae Rhodotorula mucilaginosa Sporidiobolus salmonicolor Sporobolomyces carnicolor Sporobolomyces carnicolor Sporobolomyces longuisculus Sporobolomyces oryzicola Rhodotorula bacarum Cryptococcus albidus Cryptococcus foliicola Bulleromyces albus Cryptococcus laurentii Cryptococcus nemorosus Auriculibuller fuscus Aureobasidium pullulans Zygoascus hellenicus Lipomyces lipofer Lipomyces tetrasporus Debaryomyces hansenii Debaryomyces hansenii Candida sake Candida cidri Pichia anomala Candida fermentati Hanseniaspora clermontiae Kluyveromyces lactis Kluyveromyces hubeiensis Kluyveromyces marxianus Torulaspora delbrueckii Saccharomyces cerevisiae Zygosaccharomyces florentinus Hanseniaspora uvarum Hanseniaspora uvarum Hanseniaspora meyeri Hanseniaspora optuntiae Hanseniaspora clermontiae Pichia membranifaciens Issatchenkia occidentalis Issatchenkia terricola Yarrowia lipolytica Metschnikowia audauensis Metschnikowia pulcherrima Metschnikowia fructicola Brettanomyces bruxellensis Candida stellata Candida cidri Candida bombi Pichia anomala Candida boidinii AY646097 AY646097 AY646097 AY167603 AB217494 AY167607 AY158641 AY158641 AY158657 DQ363328 AF352055 AY296054 AY557599 AF444758 AF459663 AF472635 AF444764 DQ523174 AY447018 U76533 U76527 AY167604 AY167604 AY536216 AF245402 AY296048 AY894826 AY953954 AY305673 AY325952 DQ139803 DQ466537 AY601161 U72165 DQ377648 DQ377648 AJ512458 AY267820 AY953954 DQ198965 DQ466536 DQ450883 AB197666 AJ745110 U45736 AF360542 DQ409181 AY160761 AF245402 Y15470 AB126676 AY791700 Identity (%) 97 93 93 95 93 93 99 93 93 92 94 99 97 99 99 98 99 100 97 99 95 99 92 98 99 98 93 98 100 99 98 98 100 99 93 98 99 99 95 94 99 99 97 97 98 100 100 100 96 94 100 98 156 Appl Microbiol Biotechnol (2007) 75:149–164 Appl Microbiol Biotechnol (2007) 75:149–164 Phylogenetic tree of bacteria sequences obtained during rpoB PCR-DGGE grape berries surface analyses from all grapes and vineyard studied. The numbers given in the branches are the bootstrap values after 1,000 repetitions; only significant values higher than 50% are shown. 0.05 represents the scale for the phylogenetic branches’ length. The accession number of the GenBank sequences are added in parentheses ƒFig. 2 and other species were detected. However, by the end of malolactic fermentation, only O. oeni remained in one instance and was associated with P. parvulus in the second. In this latter, the O. oeni band was the most intense when Fig. 3 Numeration on the anaerobic and facultative anaerobic Gram positive bacteria population (diamond), aerobic Gram-negative population (triangle) and anaerobic and facultative anaerobic Gram-negative bacteria population (circle), and rpoB PCR-DGGE profile obtained during the winemaking of a Merlot plot in an estate of the Pessac–Leognan appellation 157 the wine was put into barrel, 30 days after the first racking, and the first post-fermentation addition of SO2. The barrels were washed at each racking. Direct analysis with PCR-DGGE of the water used for this purpose illustrates the level of residual microorganisms on the wood. Some basidiomycetous species (Cryptococcus sp.) were detected on occasion. However, the majority of species are well known to be involved during fermentation: S. cerevisiae, non-Saccharomyces species like Candida stellata, and the spoilage yeast B. bruxellensis, which was the only one detected in all cases (Fig. 10). 158 Appl Microbiol Biotechnol (2007) 75:149–164 Fig. 4 Numeration on the anaerobic and facultative anaerobic Gram-positive bacteria population (diamond), aerobic Gram-negative population (triangle) and anaerobic and facultative anaerobic Gram-negative population (circle), and rpoB PCR-DGGE profile obtained during the winemaking of a Semillon plot in an estate of Pessac–Leognan appellation Discussion No complete microbial monitoring of the whole winemaking process, from grapes to bottled wine, has been performed to date. In this work, molecular PCR-DGGE analyses were used to study the evolution of bacteria and yeast populations, from fruit set in the vineyard to bottled red and white Bordeaux wines. One of the first applications of PCR-DGGE in wine microbiology dates from 2000, when Cocolin et al. (2000) validated a protocol for monitoring yeast species during wine fermentation. This approach can be used to monitor the fermentation of all types of wine: red (Renouf et al. 2006c), white (Renouf et al. 2005b), and Botrytis-affected wines (Mills et al. 2002). We have extended this approach to bacteria. For both yeast and bacteria, species diversity was very high on grape skins, whatever the environmental conditions (temperature, UV radiation, and agrochemical treatments). PCR-DGGE has reflected this diversity, even if all sequences could not be unequivocally identified, by Appl Microbiol Biotechnol (2007) 75:149–164 159 Fig. 5 Global survey for a Merlot wine: evolution of total yeasts population (square), on berries surface (curves on the left), and in the wine (curves on the right), NL1/LS2 PCR-DGGE profiles and its corresponding diagram. The bands extracted from the gel and sequenced are surrounded by a black square and the neighbor-joining phylogenetic tree built by comparison with the referenced sequences clustering with the most closely related species. The microbial system on grape skins consisted of numerous species unknown in wine, like Aureobasidium pullulans for the yeast and certain proteobacteria for the bacteria. These undoubtedly play an important role in the microbial consortium on the grape surfaces by producing exopolysaccharides, which can constitute a biofilm (Renouf et al. 2005a). Certain of these species, the bacteria of Serratia genus (Prem and Sriphati 2004) and the yeasts Cryptococcus sp. and Aureobasidium pullulans (Manachini et al. 1988) are also known for their pectolytic or cellulolytic activities, which can degrade the vegetables cells and provide a nutritive source. The species most commonly found in wine (S. cerevisiae, B. bruxellensis, O. oeni, etc.) were also present, but to a much lesser extent, so that their detection was sometimes only possible after specific enrichment. However, these species are an integral part of the microbial ecosystem on grape skins. The ratio between dominant and minor species on grape surfaces is problematic when conducting an inventory (Prakitchaiwattana et al. 2004). Depending on environmental conditions, the bestadapted species constitute the overwhelming majority, and the population ratio of different species can exceed 1,000fold, making the detection of minor species difficult. Indeed, in addition to the experimental variability, the DNA extraction yield can be affected by the number of cells, the wide range of cellular types, their susceptibility to glass bead disintegration, and the presence of nontargeted organisms and inhibitors. Furthermore, PCR does not allow the amplification of minor DNA sequences. However, this is not the only reason, and all possible explanations have not been fully identified. The disadvantage of PCR is its unpredictability, which can lead to the distortion of the relative abundance ratios of the original samples (Prakitchaiwattana et al. 2004). Therefore, failure to detect certain species on DGGE gel does not necessarily mean that the species is absent, only that some species are less numerous than others. To resolve this problem and reveal the presence of certain interesting species, the berries were directly placed in the enrichment medium. 160 Appl Microbiol Biotechnol (2007) 75:149–164 Fig. 6 Example of old vintage analysis by epifluorescence (picture), population enumeration on nutritive media (table), and NL1/LS2 PCR DGGE (gel) These enrichment tools must be associated with direct analysis in order not to loose any data. Many yeast species detected in grape musts were previously present on grape skins. This is true for Candida stellata, one of the most widespread species to be found on skins at harvest time (Renouf et al. 2005a), and also periodically detected on DGGE profiles during aging. This species, together with Hanseniaspora uvarum, Debaryomyces hansenii, and Pichia anomala, which are also detected on unripe grapes, is known to be active during the first phase of fermentation. The contribution of the above species can be favorable to wine quality (Lambrechts and Pretorius 2000). However, species diversity constantly decreased over time during the winemaking process. The most significant decrease was observed during alcoholic fermentation. Bacterial diversity in white wines remained higher and lasted longer than in red wines during alcoholic fermentation. This leads us to conclude that phenolic compounds may be involved in selecting species most suited to a red wine environment (Campos et al. 2003). The addition of SO2 at the end of fermentation led to another decrease in diversity. As for yeast, B. bruxellensis was undoubtedly the most resistant species. It is not very demanding from a nutritional point of view compared to other wine yeast species (Uscanga et al. 2000), and it is remarkably resistant to high ethanol concentrations (Renouf et al. 2006c). As for bacteria, levels of O. oeni and P. parvulus remained high in some instances. These species can remain in wine all throughout barrel aging as well as several years after bottling despite low nutrient concentrations and the absence of oxygen. The species most Fig. 7 Phylogenetic analysis of yeast sequences obtained by NL1/LS2 PCR DGGE of old bottles listed in Table 2 Frequency of the detection (%) 100 Saccharomyces cerevisiae (AY601161) Kluyveromyces lactis (AY305673) 63 100 5% IX Zygosaccharomyces bailii (AF399789) 97 100 Pichia anomala (AB126676) 93 100 5% 5% VIII 80 20 % IV VII 10 % III VI Brettanomyces bruxellensis (DQ409181) 100 I Rhodotorula mucilaginosa (AB217494) 92 100 0.05 5% II 72 V 5% 100 % 5% Appl Microbiol Biotechnol (2007) 75:149–164 161 Fig. 8 Phylogenetic analysis of LAB sequences obtained by rpoB PCR DGGE of old vintages listed in Table 1 Frequency of the detection (%) E 86 5% Lactobacillus plantarum (AY875849) 66 D 90 100 Pediococcus damnsosus (DQ176043) Pediococcus parvulus (AY875850) 100 5% 50% B 100 Oenococcus oeni (AY875845) 90% A Leuconostoc mesenteroides (DQ176044) 100 C 5% 0.02 detrimental to wine quality, P. parvulus and B. bruxellensis, were easily and clearly identified. This proves the usefulness of PCR-DGGE for the detection and monitoring of spoilage microbial agents. Wood is porous, and its absorbent structure allows progressive microbial penetration (Swaffield and Scott 1995; Swaffield et al. 1997), especially during the first time it is used (Renouf and Lonvaud-Funel 2005), and barrel aging is not without problems because attached microorganisms are more resistant to environmental changes and antimicrobial agents (Carpentier and Cerf 1993). These microorganisms can develop when they come into contact with wine, with a harmful effect on quality, as has been established in cider (Del Campo et al. 2003). Monitoring microbial populations on oak barrel surfaces by PCR-DGGE revealed the presence of conventional wine species: O. oeni, L. plantarum, and S. Fig. 9 Cleaning tanks water analyses by direct rpoB PCR-DGGE. The bands extracted from the gel and sequenced were surrounded by a black square and the neighbor-joining phylogenetic tree was built to compare them with the referenced sequences. For the first case (A and B): A tank cleaning after the post-fermentation maceration at devatting, B tank cleaning just after the malolactic fermentation and the post- fermenting sulphiting, for the second case (C, D, E): C tank cleaning after the post alcoholic maceration at devatting, D tank cleaning just after the malolactic fermentation and the post-fermenting sulphiting at racking moment, E tank cleaning after the second racking and barreling 30 days after the sulphiting 162 Appl Microbiol Biotechnol (2007) 75:149–164 Fig. 10 Cleaning barrels water analyses by direct NL1/LS2 PCR-DGGE. The bands extracted from the gel and sequenced were surrounded by a black square and the neighborjoining phylogenetic tree was built to compare them with the referenced sequences cerevisiae, and the spoilage species B. bruxellensis, as well as wood yeast species belonging to the Cryptococcus genera. The latter are usually not present in wine and should be extracted from wood during the washing process. Cryptococcus yeasts may also be able to decompose wood compounds and provide nutrients for other wine microorganisms (Prem and Sriphati 2004). Although direct DGGE analysis gives an exhaustive image of the species present in wine, it does not provide quantitative data. This was, therefore, completed with conventional isolation and enumeration of microbial populations. However, when comparisons are made between species identified after DNA extraction from biomass plates and DNA directly extracted from food, many species are detected only as DGGE bands in direct analysis (Ercolini et al. 2003). Therefore, this approach provides information about the percentage of noncultivable species, and conversely, this comparison can also be used to verify the selectivity of some culture media (Miambi et al. 2003) and the suitability of enrichment media (Renouf and LonvaudFunel 2006). As in other fields, conventional microbiological methods are not suitable for studying the complex microbial systems found in enology. PCR-DGGE, a culture-independent molecular method, proved to be a very sensitive tool to study the microbial community in wine and its fluctuation. Nevertheless, an approach combining the culture-independent PCR-DGGE identification method and the culturedependent numeration method is needed to investigate microbial communities on grape surfaces to detect dominant and minor species and also during winemaking when estimating populations is crucial. Regarding technical aspects, further work should be done on how to overcome DNA extraction problems in samples and on DGGE drawbacks due to experimental variability during DNA extraction, DNA matrix ratios, Appl Microbiol Biotechnol (2007) 75:149–164 biases in primer annealing (Susuki and Giovannoni 1996), base pair mistakes, and DGGE resolution to be certain that the presence and intensity of bands on DGGE gel represent changes, in all instances, of microbial diversity. References Amann R, Ludwig W (2000) Ribosomal RNA-targeted nucleic acid probes for studies in microbial ecology. FEMS Microbiol Rev 24:555–565 Ampe F, Sirvent A, Zakhia N (2001) Dynamics of the microbial community responsible for traditional sour cassava starch fermentation studied by denaturing gradient gel electrophoresis and quantitative rRNA hybridization. Int J Food Microbiol 65:45–54 Campos FM, Couto JA, Hogg TA (2003) Influence of phenolic acids on growth and inactivation of Oenococcus oeni and Lactobacillus hilgardii. J Appl Microbiol 94:167–174 Carpentier B, Cerf O (1993) Biofilms and their consequences, with particular reference to hygiene in the food industry. J Appl Bacteriol 75:499–511 Chatonnet P, Dubourdieu D, Boidron JN, Pons M (1992) The origin of ethyphenols in wines. J Sci Food Agric 60:165–178 Cocolin L, Bisson LF, Mills DA (2000) Direct profiling of the yeast dynamics in wine fermentations. FEMS Microbiol Lett 189:81– 87 Coenye T, Vandamme P (2003) Intragenomic heterogeneity between multiple 16S ribosomal RNA operons in sequenced bacterial genomes. FEMS Microbiol Lett 228:45–49 Coton E, Coton M (2003) Microbiological origin of « framboisé » in French ciders. J Inst Brew 109:299–301 Coton E, Rollan GC, Bertrand A, Lonvaud-Funel A (1998) Histamine-producing lactic acid bacteria in wines: early detection, frequency and distribution. Am J Enol Vitic 49:19–23 Del Campo G, Santos JI, Berregi I, Velasco S, Ibarburu I, Duenas MT, Irastorza A (2003) Ciders produced by two types of presses and fermented in stainless steel and wooden vats. J Inst Brew 109:342–348 Dent KC, Stephen JR, Finch-Savage WE (2004) Molecular profiling of microbial communities associated with seeds of Beta vulgaris subsp. Vulgaris (sugar beet). J Microbiol Methods 56:17–26 Ercolini D, Hill PJ, Dodd CE (2003) Bacterial community structure and location in Stilton cheese. Appl Environ Microbiol 69:3540– 3548 Felsentein J (1985) Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783–791 Guillamon JM, Sabate J, Barrio E, Cano J, Querol A (1998) Rapid identification of wine yeast species based on RFLP analysis of the ribosomal ITS regions. Arch Microbiol 169:387–392 Kumar S, Tanura K, Jakobsen IB, Masatoshi N (2001) MEGA2: Molecular Evolutionary Genetics Analysis software, Arizona State University, Tempe, AZ USA Kurtzman CP, Robnett CJ (1997) Identification of clinically important ascomycetous yeasts based on nucleotide divergence in the 5′ end of the large-subunit (26S) ribosomal DNA gene. J Clin Microbiol 35:1216–1223 Lambrechts MG, Pretorius IS (2000) Yeast and its importance to wine aroma—a review. S Afr J Enol Vitic 21:97–129 Le Jeune C, Lonvaud-Funel A (1994) Lactobacillus hilgardii and Lactobacillus brevis DNA analysis by restriction fragment length polymorphism. Food Microbiol 11:195–202 Lonvaud-Funel A, Joyeux A, Ledoux O (1991) Specific enumeration of lactic acid bacteria in fermenting grape must and wine by 163 colony hybridization with non-isotopic DNA-probes. J Appl Microbiol 71:501–508 Lopez I, Ruiz-Larrea F, Cocolin L, Orr E, Phister T, Marshall M, Vanderghenst J, Mills DA (2003) Design and evaluation of PCR primers for analysis of bacterial populations in wine by denaturing gradient gel electrophoresis. Appl Environ Microbiol 69:6801–6807 Manachini PL, Parini C, Fortina MG (1988) Pectic enzymes from Aureobasidium pullulans LV 10. Enzyme Microb Technol 10:682–685 Miambi E, Guyot JP, Ampe F (2003) Identification, isolation and quantification of representative from fermented cassava dough using an integrated approach of culture-dependent and cultureindependent methods. Int J Food Microbiol 82:111–120 Millet V, Lonvaud-Funel A (2000) The viable but non-culturable state of wine micro-organisms during storage. Lett Appl Microbiol 30:136–141 Mills DA, Johansen EA, Cocolin L (2002) Yeast diversity and persistence in Botrytis-affected wine fermentations. Appl Environ Microbiol 68:4884–4893 Muyzer G, de Waal EC, Uitterlinden AG (1993) Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction amplified genes coding for 16S rRNA. Appl Environ Microbiol 59:695–700 Prakitchaiwattana CJ, Flett GH, Heard GM (2004) Application and evaluation of denaturing gradient gel electrophoresis to analyze the yeast ecology of wine grapes. FEMS Yeast Res 4:865–877 Prem AA, Sriphati K (2004) Digestion of cellulose and xylan by symbiotic bacteria in the intestine of the Indian flying fox (Pteropus giganteus). Comp Biochem Physiol 139:65–69 Rantsiou K, Comi G, Cocolin L (2004) The rpoB gene as a target for PCR-DGGE analysis to follow lactic acid bacterial population dynamics during food fermentations. Food Microbiol 21:481– 487 Renouf V, Lonvaud-Funel A (2005) Incidence microbiologique de l’usage de barriques neuves et/ou de barriques usagées. Rev Fr Oenol 211:10–14 Renouf V, Lonvaud-Funel A (2006) Development of an enrichment medium to detect Dekkera/Brettanomyces bruxellensis, a spoilage wine yeast, on the surface of grape berries. Microbiol Res (in press) Renouf V, Claisse O, Lonvaud-Funel A (2005a) Numeration, identification and understanding of microbial biofilm on grape berry surface. Aust J Grape Wine Res 11:316–327 Renouf V, Claisse O, Lonvaud-Funel A (2005b) Use of the rpoB gene as a target for PCR-DGGE analysis to follow bacteria species evolution on the grape berry surface and during the red and white winemaking process. 8th Symposium on lactic acid bacteria. Amsterdam, The Netherlands, 25–28 June Renouf V, Claisse O, Lonvaud-Funel A (2006a) rpoB gene: a target for identification of LAB cocci by PCR-DGGE and melting curves analyses in real time PCR. J Microbiol Methods 67:162– 170 Renouf V, Claisse O, Miot-Sertier C, Lonvaud-Funel A (2006b) LAB evolution during winemaking use of rpoB gene as a target for PCR-DGGE analysis. Food Microbiol 23:36–145 Renouf V, Falcou M, Miot-Sertier C, Perello MC, de Revel G, Lonvaud-Funel A (2006c) Interactions between Brettanomyces bruxellensis and the other yeasts species during the first steps of winemaking. J Appl Microbiol 100:1208–1219 Saitou N, Nei M (1987) The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4:406–425 Sigler WV, Miniaci C, Zeyer J (2004) Electrophoresis time impacts the denaturing gradient gel electrophoresis-based assessment 164 of bacterial community structure. J Microbiol Methods 57: 17–22 Susuki MT, Giovannoni SJ (1996) Bias causes by template annealing in the amplification of mixtures of 16 rRNA genes by PCR. Appl Environ Microbiol 62:625–690 Swaffield CH, Scott JA (1995) Existence and development of natural microbial populations in wooden vats used for alcoholic cider maturation. J Am Soc Brew Chem 53:117–120 Swaffield CH, Scott JA, Jarvis B (1997) Observation on the microbial ecology of traditional alcoholic cider storage vats. Food Microbiol 14:353–361 Appl Microbiol Biotechnol (2007) 75:149–164 Uscanga MGA, Delia ML, Strehaiano P (2000) Nutritional requirements of Brettanomyces bruxellensis: growth and physiology in batch and chemostat cultures. Can J Microbiol 46: 1046–1050 Uthurry CA, Suarez Lepe JA, Lombardero L, Garcia Del Hierro JR (2005) Ethyl carbamate production by selected yeasts and lactic acid bacteria in red wine. Food Chem 94:262–270 Walling E, Gindreau E, Lonvaud-Funel A (2001). La biosynthèse d’exopolysaccharides par des souches de Pediococcus damnosus isolées du vin: mise au point d’outils moléculaire de détection. Lait 81:289–300